Health technology assessment methods for evaluating medical tests: developing a novel application of the linked evidence approach

Abstract

BACKGROUND The health consequences of medical testing are often not apparent or easily measured. To address this, the ‘linked evidence approach’ (LEA) was developed to estimate the clinical utility of a test so that policy makers can make informed public funding decisions. Australia has the largest international experience with the application of LEA. RESEARCH AIM 1 The first aim of the presented research was to investigate the feasibility, utility and policy impact of LEA. To enable the use of LEA in test evaluation there needed to be a more rigorous approach taken to determine the risk of bias in test accuracy studies. An existing evidence hierarchy recommended by the Australian Government for use in health technology assessment (HTA) was consequently revised between 2005 and 2009 to consider design-related biases in test accuracy studies. The hierarchy underwent a national public consultation and pilot process and became widely used. A study was conducted to model the overall impact of LEA on health policy; data were extracted from HTA reports commissioned before-and-after the use of LEA was mandated by the Australian Government in 2005. Logistic regression analyses and regression diagnostics were performed to estimate model fit, model specification and to inform model selection. There was no discernible impact of LEA on the direction of public funding decisions (OR=1.36, 95%CI 0.62, 3.01) but the use of LEA did strongly predict that a medical test would not receive interim funding (Χ²=12.63, df=1, p=0.0004). This suggests that the method enables greater certainty in decision-making. RESEARCH AIM 2 The second aim was to develop guidance on how LEA should be applied during the evaluation of medical tests. A systematic literature review was performed on the methods used in HTAs evaluating medical tests so that a decision framework could be constructed to guide the application of LEA and to address potential methodological problems with the approach. The framework systematises the application of LEA by categorising medical tests into three possible scenarios, namely optimisation, trade-off and disease-spectrum change. The evidence collation and linkage practices need to be tailored to each of these scenarios. RESEARCH AIM 3 The final aim of the presented research was to adapt LEA to the evaluation of a drug and its companion diagnostic test (‘personalised medicine’). An analysis of guidance documents and a review of case studies was undertaken to identify key information to guide decisions concerning the reimbursement of personalised medicines. An evaluation framework, incorporating LEA, was created to determine the safety, effectiveness and cost-effectiveness of personalised medicines. 79 evaluation items were proposed and examples provided to demonstrate the linkage of different types of evidence to reduce decision-maker uncertainty. The framework underwent a public consultation and pilot process. The impact of the evaluation framework on public funding decisions was critically reviewed in the three years’ after the framework was implemented nationally. CONCLUSIONS This thesis by publication resulted in three theoretical methods papers (published), one analytical paper (under review) and one published review paper (invited). The methods developed for these publications were aimed at improving how medical tests are considered and valued by our health systems. LEA enables the clinical utility of medical tests to be estimated, leading to greater certainty for policy makers and reducing the need for ‘interim’ funding decisions. Methods for standardising the application of LEA have allowed consistent information to be provided to policy makers. The adaptation of LEA to the evaluation of personalised medicines has enabled previously siloed funding decisions on companion tests and therapeutics to be integrated. The research outputs from this thesis have directly affected technology evaluation practice, with consequent impacts on health policy and test subsidy decisions.